Upcoming Seagate drives will reach 16TB

As the new and incredibly-fast kids on the block, solid-state drives get a lot of attention these days—particularly as they get large enough to handle the average user's everyday needs without the services of a second drive. However, there are many usage scenarios where a drive's size matters more than its speed. For these applications, Seagate has new high-capacity hard drives coming to market relatively soon, with the largest of them reaching 16TB.

In an earnings call this week, Seagate's CEO Steve Luczo revealed that the company will release those drives in the next 12 to 18 months. Seagate also has plans to expand the number of different capacity points that it offers its customers. Today, its highest-capacity drive is a 10TB model, but Luczo revealed that Seagate already has a 12TB product in customers' hands for evaluation. The drive seems to be performing up to standards, seeing as the company has plans for releasing 14TB and 16TB models.

Luczo indicated that demand for higher-capacity hard drives is coming from several different markets. Companies deploying cloud-based storage have been good customers for Seagate, but the consumer market has also been growing. Specifically, Seagate points out that consumers are looking for high-capacity drives for surveillance and network-attached storage systems.

Looking forward, Luczo mentioned 20TB drives as a possibility, but primarily to remind its investors that creating these drives has and will require a significant investment of time and resources. It's impressive that a drive of that size is even part of the conversation, however, considering that we were just speculating about drives hitting these capacities a mere five years ago.

Comments closed
    • Chrispy_
    • 3 years ago

    Whenever Seagate talks about high capacity drives, I always assume that they’ve done it by turning up the SMR ‘compression’ even further, at the cost of random massive write performance degredation on anything other than nice, easy sequential streaming.

    That’s why you see them advertised as “archive” and “surveillance” because they absolutely blow chunks when it comes to typical consumer usage patterns.

    If Seagate can make a 16TB drive without SMR, I’ll start listening to what they have to say.

      • Waco
      • 3 years ago

      Typical consumer usage patterns are [i<]faster[/i<] on SMR drives due to the log-structured initial write.

      • not_a_gerbil
      • 3 years ago

      “because they absolutely blow chunks when it comes to typical consumer usage patterns.”

      Boot drives should be SSD. Beyond 1TB, I really, really, doubt that consumers work with active datasets larger than 1TB. My hobby is capturing/filtering/encoding HD tv to mp4. (someone has to do the work that eventually shows up on torrents) My active dataset (a season full of HD shows being processed by avisynth/megui) is never over 200GB.

      Consumers use large drives for archives. Its photos, videos and music. None of that needs fast random access. You write it once and read it often. Shingled is fine for reads. Its writes that are very bad.

      I buy HGST because of reliability. But if HGST had a shingled product that was cheap $/TB, I’d buy it without hesitation for my media server (currently 8TB filled out of 12TB).

        • Chrispy_
        • 3 years ago

        Ridiculously simple, ordinary tasks grind to a halt on shingled drivers. Updating indexes, updating thumbnails, writing to the disk when it’s fragmented (even slightly) doing two things at once, like downloading torrents whilst installing a game.

        Blows [i<]chunks[/i<]. For sequential backup streams or recording a single video stream they're absolutely fine. But for the love of god don't try to use them for more than one sequential thing or EVER let them get even slightly fragmented.

          • Waco
          • 3 years ago

          Funny, I do parallel writes to them daily without issue.

    • Fonbu
    • 3 years ago

    Having such a large capacity drive is dangerous, but I remember people years ago saying that about 1TB hard drives and look where we are today. Those really large drives need a good warranty and more reliability features. Its almost practical to include a data recovery service voucher when you purchase such a large drive, because you know the inevitable can and will happen. That is my two cents.

      • BillyBuerger
      • 3 years ago

      Not really. Any size drive that has important data on it requires a backup strategy. Loosing a 16GB flash drive with important information on it can be just as bad as loosing 16TB of data. If the data is important, then for a simple example, you always buy two. If you currently have a 1TB drive with data you don’t want to loose, have a second that you regularly backup to. If you have a 16TB drive, then have a second that you regularly backup to. No real difference.

      And I would say that just because we are using more space doesn’t necessarily mean we have more data. Movies have moved from low quality VHS which you could store at quality in a very small file. A movie today that’s the same length fills up a Blu Ray drive at like 50GB. Same amount of information but uses a lot more space due to increased quality. These larger drives just continue that.

        • Fonbu
        • 3 years ago

        That is excellent explanation!

        • dragontamer5788
        • 3 years ago

        [quote<]If you currently have a 1TB drive with data you don't want to loose, have a second that you regularly backup to. If you have a 16TB drive, then have a second that you regularly backup to. No real difference.[/quote<] While this is true, it should be noted that a 1TB drive from a few years ago may have a speed of ~120MB/s. The 16TB drive today will have what? 200MB/s in the best case? (PMR + Helium technology) and maybe 20MB/s write speed in the worst case? (If Shingled). 200MB/s means it will take a full day to mirror a 16TB drive. I mean yeah, you've got RAID arrays and all that, but each time a RAID array gets messed up, you'll have to do a full rebuild of data which means ~24+ hours out of commission (in the 200MB/s case!). If you want to actually finish a rebuild within a singular workday, you need to have ~200MB/s in 5ish hours, or ~3TB or 4TB drives.

          • Pwnstar
          • 3 years ago

          This is why any RAID past RAID 1 is a mistake. Just mirror your drives and you’ll be fine.

            • Waco
            • 3 years ago

            Please explain how you recover from a single error reading from that single drive when your second drive dies. 🙂

            • joselillo_25
            • 3 years ago

            single bite error in drives are rare, but much more difficult to solve in a RAID setup than in a single drive, specially for all those people that use RAID NAS as a backup

            do not argue on this, RAID on desktop is ridiculous and is just a way to complicate your life. Even just mirror drive in a USB from time to time is safer than use a RAID in a desktop enviroment.

            I have helped 2 friends to recover drives from a NAS RAID dissaster and is a no-go in a desktop, specially know that even windows10 can pool drives and create storage spaces pretty easy.

            You will know when the dissaster happens.

            • Waco
            • 3 years ago

            Wut?

            • dragontamer5788
            • 3 years ago

            You made the mistake of making a very poorly worded post addressing a very obscure, minor issue in the parent post… arguing over a point that is rather technical.

            [quote<] do not argue on this, RAID on desktop is ridiculous and is just a way to complicate your life. Even just mirror drive in a USB from time to time is safer than use a RAID in a desktop enviroment. I have helped 2 friends to recover drives from a NAS RAID dissaster and is a no-go in a desktop, specially know that even windows10 can pool drives and create storage spaces pretty easy.[/quote<] The problem here is that hardware RAID sucks (even "soft-hardware RAID", like what's on AMD / Intel Motherboards which is mostly a firmware hack). For all practical purposes, "RAID" is a messed up, old technology that doesn't solve most issues. "Full Software Raid" solutions are much easier to use. That's "Storage Spaces" on Windows, and "Unraid" on Linux, "ZFS" for BSD. I mean, yeah. RAID is dead for all practical purposes. A NAS should be using ZFS or Storage Spaces. Desktops should be using Storage Spaces. [b<]Period.[/b<] So you're correct here. --------------- [quote<] single bite error in drives are rare, but much more difficult to solve in a RAID setup than in a single drive, specially for all those people that use RAID NAS as a backup[/quote<] This is true, and joselillo_25 seems to be speaking from experience. RAID offers no protection against single-bit errors or "bit-rot". You are required to have higher-level checksums (such as those offered in "Storage Spaces" or "ZFS") to detect bit-rot scenarios. Most "RAID" implementations only redundantly writes data out. It doesn't actually do any error-checking when reading data. As such, if a bit flips silently (ie: the bitrot scenario. When a bit flips while the computer is off), RAID just doesn't have any tools to handle the scenario.

            • Pwnstar
            • 3 years ago

            Yup, that’s why I use ZFS.

            • Pwnstar
            • 3 years ago

            I’ve been doing this for 15 years now and never have had a problem. Probably because I use double backups.

      • just brew it!
      • 3 years ago

      The primary customers for these types of drives are going to be deploying them in storage arrays that provide some form of redundancy.

    • Krogoth
    • 3 years ago

    I’m willing to bet that these new drives are going to be using SMR and are going to be aimed for datacenter customers like the current corp of 8TB+ HDDs HDDs.

      • Vhalidictes
      • 3 years ago

      And if they are SMR they will be useless for consumer use. You think spinning rust lags SSD performance NOW…

        • dragontamer5788
        • 3 years ago

        SMR is just slow on the write speeds. It should be fine with read speeds, which is more typical for NAS usage.

        EDIT: I just thought about drive-rebuilds in RAIDish (or ZFS / Storage Space / etc. etc.) . Rebuilding would be at 15MB/s or so, for 8TB? That’s [b<]6 days[/b<] to rebuild. So you actually don't want to use SMR for RAID.

        • meerkt
        • 3 years ago

        For bulk storage, which is supposedly what you’d do on HDDs vs SSDs, I think it should be fine. Not that I’ve tried.

          • Chrispy_
          • 3 years ago

          I’ve tried and I won’t try again.

          They’re called “archive” and “surveillance” drives specifically because Seagate can’t get away with calling them desktop drives. They’re not even good for storing your photos and videos on.

            • Takeshi7
            • 3 years ago

            Surveillance drives aren’t SMR. surveillance is the exact opposite of a use case that would be good for SMR.

            • Waco
            • 3 years ago

            Disagree. Host managed drives handle this perfectly.

            • Takeshi7
            • 3 years ago

            Is there a surveillance system that supports host managed SMR? I don’t know of any.

            • Waco
            • 3 years ago

            No, I’m just saying that SMR drives are totally capable of this type of workload (quite easily, really), given a host that is aware of how to talk to them properly.

            • Waco
            • 3 years ago

            I think you’re doing the wrong things to your storage if you’re having problems with them.

            • JustAnEngineer
            • 3 years ago

            I tried a 3 TB Seagate drive in my TiVo Bolt. The terrible SMR performance caused video glitches when the buffer couldn’t keep up with two shows recording at once, then it corrupted the file system and caused a system failure. Of course, I didn’t realize that this particular drive used Seagate’s horrifically bad SMR technology when I ordered it because Seagate intentionally obfuscates this information. It didn’t appear on their datasheet for this drive nor in the information that the vendor provided. I only found it in later 3rd party reviews after I had installed the drive.

            As a consequence of SMR’s failure, I permanently lost a couple hundred old recordings that I had transferred from my previous TiVo box to the new one, I had no TV for a week and I spent 2½ hours on the phone with support for TiVo and Comcast. If I had tried to use a Seagate SMR drive in a more critical service, the consequences would have been more severe.

            I’ve decided that the easiest way for me to avoid losing more data and wasting more time is to avoid buying products from Seagate until they abandon SMR or get a more ethical marketing group in place.

            • Waco
            • 3 years ago

            I didn’t know they made a 3 TB SMR.

            They definitely do hide that information more than they should though.

            EDIT: You sure it was SMR? What was the model number? As far as I know, the smallest SMR in 3.5″ is 5 TB.

            • not_a_gerbil
            • 3 years ago

            “They’re not even good for storing your photos and videos on.”

            Why do you say that? They can read at 90MB/s. 4K Bluray on 100GB discs (not even released yet) is only 16MB/s.

            You can’t torrent to the drive or have 10 people reading at once but otherwise it should be fine.

            • dragontamer5788
            • 3 years ago

            Actually, since NAS has a very large “sustained use” scenario (RAID rebuilds), I’m thinking that this would be a bad drive for NAS or other multi-hard drive environments.

            For single-drive environments, SMR has very good burst performance (as long as the data to-be-written fits inside of the buffer, its just as fast as PMR). So SMR would be better at typical consumer usage scenarios, but bad at “archival purposes”. (Where “archive” is defined as NAS)

            • Waco
            • 3 years ago

            RAID rebuilds are actually the perfect workload for them, if they recognized streaming writes from LBA 0 -> end of drive properly.

            I think whether that happens depends on a LOT of variables not always in control of the consumer.

        • BIF
        • 3 years ago

        You made me think of “green” drives, which I hate with a passion.

        They’re so slow, I always think my system is broken.

          • HERETIC
          • 3 years ago

          Horses for courses-
          Sounds like you’ve had problems with the early WD greens.
          Remember a 1TB one that chose 30MB as it’s happy write speed.
          Today most slower RPM drives are more like 150MB(outside platter)
          to 75MB(inside platter) good enough on average to saturate network.

    • ronch
    • 3 years ago

    Unless i download torrents left and right all day long I’m not sure I’d need more than 2-3TB.

      • TheEldest
      • 3 years ago

      You … you do realize non-consumers use lots of drives, right? Like facebook. Facebook needs quite a lot of capacity for user uploaded photos.

      Crashplan for users’ backups or backblaze.

      So while YOU may not need more than 2-3 TB, lots of other people definitely need the larger capacities.

        • ronch
        • 3 years ago

        NOT speaking for EVERYONE, as I mentioned.

      • jihadjoe
      • 3 years ago

      A non-pirate home theater enthusiast with a Network Receiver and a NAS/Media Server could fill that up pretty quickly. A couple of hundred Blurays worth of movies and TV shows at 25-50GB each and you’re easily past 10TB. And that’s before we go into UHD Blurays which can be up to 128GB each.

        • shalmon
        • 3 years ago

        i’m at exactly that now.. .10tb mirrored (two 4tb pairs and one 2tb pair) plus a 256gb ssd for my system drive

        and i don’t have a ton of storage left…

        mind you my woman’s files take up a lot of space and i’d bet she’s on the abnormaly high side of space usage when it comes to spousal storage space requirements.. at least when compared to the rest of my friends and family

        • shaq_mobile
        • 3 years ago

        Yeah directly ripping br without significant compression is what… 25 to 40gb per disc. The average well compressed 1080 mkv is maybe 8gb but probably closer to 5.

        The amount of reasonable ways to view shows has, in my mind, largely undermined the drive for piracy. Movies are still a bit tough, but I really think that the market has reasonably responded to the underlying cause for most piracy. Surprisingly, you can still find Blu-ray for sale at the local grocery for $30.

      • Ifalna
      • 3 years ago

      As a consumer, the only thing that will manage to break that 3 TB barrier (and rather easily) is video data. Esp barely compressed high resolution video data slated for further processing.

      Doesn’t have to be pirated stuff though. Family trip recorded in 4K? Yikes.
      Recorded game sessions at high bitrate? Ouch.

        • JustAnEngineer
        • 3 years ago

        Each photo that I shoot takes up about 30 MB. Shoot hundreds of photos per day on that dream vacation and the gigabytes add up.

        • Krogoth
        • 3 years ago

        [quote<]Family trip recorded in 4K? Yikes. Recorded game sessions at high bitrate? Ouch. [/quote<] Only videophiles put forth the extra effort. The mainstream crowd is going to be opting for 720p/1080p.

    • adisor19
    • 3 years ago

    Bring back 5 year warranty and I might consider it.

    Adi

      • Takeshi7
      • 3 years ago

      They do have a 5 year warranty on their higher end models.

      • NTMBK
      • 3 years ago

      Stop redundantly signing your posts and we might consider it.

      -Seagate

        • ronch
        • 3 years ago

        Yeah. Why such formality?

        -ronch

          • fhohj
          • 3 years ago

          Hey, I do a thing!

          -Potato

        • Wirko
        • 3 years ago

        Then again, Seagate would be extremely pleased if every forum poster included a 25-kilobyte signature in every single post.

    • Alexko
    • 3 years ago

    How many platters are we talking about, here?

      • Growler
      • 3 years ago

      All. All of the platters.

      • adisor19
      • 3 years ago

      Yes.

      • morphine
      • 3 years ago

      You need to think in terms of serving trays. Upgrades, yo.

      • meerkt
      • 3 years ago

      It’s probably slightly higher density than already announced drives, so I’m guessing 8 platters, and Helium-based:
      [url<]https://www.extremetech.com/computing/240878-western-digital-announces-new-high-capacity-12tb-14tb-helium-filled-hard-drives[/url<]

      • albundy
      • 3 years ago

      dont they come in a set of 12? they gotta match the bowls, right?

    • chuckula
    • 3 years ago

    Using an unrealistically best-case scenario of a continuous 200MB/sec transfer to the drive, it would take over 22 hours to write the drive in full.

    Being more realistic and pegging the write speed at 90MB/sec or so, it’s just over 2 days.

      • PrincipalSkinner
      • 3 years ago

      You better act fact once you see it started crapping out.

      • BillyBuerger
      • 3 years ago

      Yeah, I thought about saying that. But most backups solutions do differential backups. So it only copies what’s changed. Initial seed would take a long time yes. And if you for some reason move or replace everything on your drive, then yes. But under normal situations, that shouldn’t be an issue. But it is something to think about. Whereas platter drives are still the cheapest way to store large amounts of data, the speed isn’t increasing as fast as the density. So the issue is getting worse in that regard.

        • BIF
        • 3 years ago

        Defragging is done less often now with so many people using SSDs, but it’s good to note that any files moved by a defragger will get backed up in the next incremental or differential. That could make for an unacceptably long incremental backup time to completion and/or for an unacceptably large file size. Multiply by every partition being defraged-and-backed-up, and you could fill your backup drives sooner than originally expected.

        To cut down on needlessly oversized incremental backup file sizes (and the time needed to back them up/copy them to an offsite location, etc), just schedule defrags to run just prior to the next full backup. The full backup will back up everything anyway and will then reset the last-used indicator for all files, and you’re done.

        I did this more often when I had a lot of spinning drives, but now I only use HDDs for backups and for download folders, office data files, and other low-priority data that doesn’t require SSD speed. My SSDs only get TRIM’d. The defrag issue is mostly a non-issue for me now.

Pin It on Pinterest

Share This